mlr3 1.3.0

• feat: Learner$predict() can now add additional data to PredictionClassif and PredictionRegr objects via the extra field.
• feat: Mlr3Error and Mlr3Warning classes for errors and warnings.
• refactor: $obs_loss methods in Measure, Prediction, ResampleResult, and BenchmarkResult.
• perf: Use more lengths() and data.table::setattr() in tests.
• perf: Use attr() instead of attributes() for extracting single attributes.
• test: Use more specialised test functions.
• fix: weights_measure now work with stratum.
• fix: Encapsulation loads loaded packages on the workers.
• fix: Learners can handle new factor levels.
• refactor: Use more implicit returns.
• feat: Learner can pass extra data to predictions.

mlr3 1.2.0

• feat: Add mirai support for parallelization and encapsulation.
• feat: Fallback can now be configured to only be used in case of certain errors via the when argument.
• feat: Custom error and warning classes.
• fix: $selected_features returns error when model is not trained yet.
• docs: Missing values during scoring.
• BREAKING CHANGE: Removed data_format argument of $data() method of DataBackend.
• BREAKING CHANGE: Remove data_formats field from Learner.
• BREAKING CHANGE: Remove DataBackendMatrix class.
• feat: Add materialize_view() method to Task to replace the internal DataBackend with a new one after operations like $select() and $filter().
• docs: Information about quantile prediction.
• perf: Use fget in assert_predictable.
• feat: Store oob error in state without requiring storing the model.
• fix: $levels() of Task returns in the correct order.
• chore: Only print up to 10 classes in the Task printer.
• fix: Check if quantiles and quantile_response are set.

mlr3 1.1.0

• feat: Add new measure MeasureRegrRQR for quantile regression.
• feat: Add $predict_newdata_fast() method to Learner to speed up prediction.
• fix: configure_learner is passed on run_experiment() for autotest learners.

mlr3 1.0.1

• fix: The printer of Learner failed when the validate field was set.
• fix: Avoid printing empty line for feature less tasks.
• perf: Use data.table::setattr() for less copying.

mlr3 1.0.0

• BREAKING CHANGE: The mlr3 ecosystem has a base logger now which is named mlr3.
  The mlr3/core logger is a child of the mlr3 logger and is used for logging messages from the mlr3 package.
  Some extension packages have their own loggers which are children of the mlr3 logger e.g. mlr3/mlr3pipelines and mlr3/bbotk for tuning.

• BREAKING CHANGE: weights property and functionality is split into weights_learner and weights_measure:

  • weights_learner: Weights used during training by the Learner.
  • weights_measure: Weights used during scoring predictions via measures.

  Each of these can be disabled via the new field use_weights in Learner and Measure objects.

• feat: Add $confusion_weighted field to PredictionClassif.

• feat: Add $weights field to Prediction. It contains the weights_measure weights from the Task that was used for prediction.

• feat: Add "macro_weighted" option to Measure$average field.

• feat: MeasureRegrRSQ and MeasureClassifCost gain "weights" property.

• feat: LearnerClassifFeatureless, LearnerRegrFeatureless, LearnerClassifDebug, LearnerRegrDebug gain "weights" property.

• feat: Learner printer now prints information about encapsulation and weights use.

• feat: Add score_roc_measures() to score a prediction on various roc measures.

• feat: A better error message is thrown, which often happens when incorrectly configuring the validate field
  of a GraphLearner

• feat: Added method $set_threshold() to BenchmarkResult and ResamplingResult, which allows to set the threshold for the response prediction of classification learners, given they have output a probability prediction (#1270).

• feat: Added field $uhash_table to BenchmarkResult and functions uhash() and uhashes()
  to easily compute uhashes for given learner, task, or resampling ids (#1270).

• feat: You can now change the default predict type of classification learners to "prob" by setting
  the option mlr3.prob_as_default to TRUE (#1273).

• feat: benchmark_grid() will now throw a warning if you mix different predict types in the
  design (#1273).

• feat: Converting a BenchmarkResult to a data.table now includes the task_id, learner_id, and resampling_id columns (#1275).

• fix: Add missing parameters for "regr.pinball" and "sim.phi" measures.

mlr3 0.23.0

• feat: Add new col_role offset in Task and offset Learner property.
  A warning is produced if a learner that doesn't support offsets is trained with a task that has an offset column.
• fix: The $predict_newdata() method of Learner now automatically conducts type conversions (#685).
• BREAKING_CHANGE: Predicting on a Task with the wrong column information is now an error and not a warning.
• Column names with UTF-8 characters are now allowed by default.
  The option mlr3.allow_utf8_names is removed.
• BREAKING CHANGE: Learner$predict_types is read-only now.
• docs: Clear up behavior of Learner$predict_type after training.
• feat: Add callbacks to resample() and benchmark().
• fix: Internal tuning and validation now works when the model requires marshaling (#1256).

mlr3 0.22.1

• fix: Extend assert_measure() with checks for trained models in assert_scorable().

mlr3 0.22.0

• fix: Quantiles must not ascend with probabilities.
• refactor: Replace tsk("boston_housing") with tsk("california_housing").
• feat: Require unique learner ids in benchmark_grid().
• BREAKING CHANGE: Remove $loglik() method from all learners.
• fix: Ignore future.globals.maxSize when future::plan("sequential") is used.
• feat: Add $characteristics field to Task to store additional information.